CVE-2021-37665: TensorFlow GPU - Incomplete Validation in MKL Implementation of Requantization

Background

TensorFlow is an end-to-end open source platform for machine learning. It provides a comprehensive ecosystem of tools, libraries, and resources for developing and deploying machine learning models. TensorFlow GPU is a variant of TensorFlow that leverages the power of GPUs for accelerated computation. It is widely used in various industries and research domains for training and inference of deep learning models.

Vulnerability Detail

In affected versions of TensorFlow GPU, there is an incomplete validation issue in the MKL (Math Kernel Library) implementation of requantization. This vulnerability allows an attacker to trigger undefined behavior by binding a reference to a null pointer or accessing data outside the bounds of heap allocated arrays. The issue arises due to the lack of dimension validation in the input tensor of the MKL implementation. A similar issue is also present in the MklRequantizePerChannelOp implementation. Exploiting this vulnerability can lead to crashes, memory corruption, or potentially allow an attacker to execute arbitrary code.

pip install --upgrade "tensorflow-gpu==2.12.0"

Workaround

As a temporary workaround, users can consider implementing input validation checks in their own code to ensure the dimensions of the input tensor are valid before passing it to the vulnerable MKL operations. Here's an example of how to perform input validation in Python:

import tensorflow as tf

def validate_input(input_tensor):
    if input_tensor.shape[0] <= 0:
        raise ValueError('Invalid input tensor dimensions')
    # Perform other necessary validation checks

# Usage
input_tensor = ...
validate_input(input_tensor)
# Call the vulnerable MKL operation

Conclusion

In conclusion, TensorFlow GPU versions prior to the patched releases are vulnerable to an incomplete validation issue in the MKL implementation of requantization. This vulnerability can have serious consequences, including crashes, memory corruption, and potential code execution. Users are strongly advised to update their TensorFlow GPU installations to the patched versions once they are available. If you need further assistance or support in addressing this vulnerability, please reach out to DevNack.com.

See every vulnerability recorded for tensorflow-gpu →

FAQ